In this article a Kalman Filter-based disturbance observer is proposed for treating the fault detection and isolation problem of the VSC-HVDC transmission system. By proving differential flatness properties for the state-space model of the VSC-HVDC system which consists of an AC/DC converter-(rectifier), a high-voltage DC transmission line and a DC/AC inverter it is confirmed that this system can be transformed into the input-output linearized form and in the canonical Brunovsky form. The effects of model uncertainty and external perturbations are modeled as additive disturbance inputs. Moreover, by considering the disturbance inputs as additional state variables an extended state space description is obtained which is proven to be observable. For the latter state-space model a Kalman Filter-based disturbance observer is designed which is capable of identifying simultaneously both the non-measurable state variables and the disturbance terms. Next, the residuals of the Kalman Filter undergo statistical signal processing which finally enables fault detection and isolation. The sum of the squares of the residuals' sequence weighted by the inverse of the associated covariance matrix is a stochastic variable or "statistical test" which follows the $\chi^2$ distribution. The confidence intervals of the $\chi^2$ distribution allow to define fault thresholds. By comparing the "statistical test" against the fault thresholds one can conclude with a high level of confidence (of the order of 95\% and beyond) the existence of a failure. The disturbance estimates which are provided by the Kalman Filter-based disturbance observer allow also to perform fault isolation.